A portion of this work was presented at the 50th Annual Meeting of the American Society of Hematology held December 6–9th, 2008 in San Francisco, CA, USA.
Susan R. Kahn, Center for Clinical Epidemiology and Community Studies, Jewish General Hospital, 3755 Côte Ste-Catherine, Rm. A-127, Montreal, Quebec H3T 1E2, Canada. Tel.: +1 514 340 7587; fax: +1 514 340 7564. E-mail: email@example.com
Summary. Background: The pathophysiology of post-thrombotic syndrome (PTS) is postulated to involve persistent venous obstruction and venous valvular reflux. Objective: To study the association between D-dimer level, valvular reflux and the PTS in a well-defined cohort of deep vein thrombosis (DVT) patients. Methods: Consecutive patients with acute symptomatic DVT were recruited at eight centers and were followed for 24 months. D-dimer was measured at 4 months. A standardized ultrasound assessment for popliteal valvular reflux was performed at 12 months. Using the Villalta scale, patients were assessed for PTS during follow-up by evaluators who were unaware of D-dimer or reflux results. Results: Three hundred and eighty-seven patients were recruited; of these, 305 provided blood samples for D-dimer and 233 had a 12-month reflux assessment. PTS developed in 45.1% of subjects. Mean D-dimer was significantly higher in patients with vs. without PTS (712.0 vs. 444.0 μg L−1; P = 0.02). In logistic regression analyses adjusted for warfarin use at the time of D-dimer determination and risk factors for PTS, D-dimer level significantly predicted PTS (P = 0.03); when stratifying for warfarin use at the time of blood draw, adjusted odds ratio (OR) for developing PTS per unit difference in log D-dimer was 2.33 (95% CI 0.89, 6.10) in those not on warfarin vs. 1.25 (95% CI 0.87, 1.79) in those on warfarin. Ipsilateral reflux was more frequent in patients with moderate-to-severe PTS than in patients with mild PTS (65% vs. 40%, respectively; P = 0.01) and was independently associated with moderate-to-severe PTS in logistic regression analyses (P = 0.01). Conclusion: D-dimer levels, measured 4 months after DVT in patients not on warfarin, are associated with subsequent development of PTS. Venous valvular reflux is associated with moderate-to-severe PTS.
Post-thrombotic syndrome (PTS) occurs frequently after deep venous thrombosis (DVT), with a cumulative incidence of 20–50% within 2 years [1–3]. PTS is difficult to treat, leads to important morbidity and incurs high costs [1–3]. Patients who develop PTS report heaviness, pain, cramps, pruritis and paresthesias in the affected limb. Clinical examination reveals, edema, induration, hyperpigmentation, venous ectasia, erythema, pain during calf compression and in severe cases, leg ulceration [2,3]. While risk factors for PTS reported in the literature include advanced age, poor quality of anticoagulation, recurrence of DVT, proximal DVT, lack of resolution of initial DVT symptoms and signs, and elevated body mass index (BMI) [1,4,5], overall, the ability to predict which patients with DVT are likely to develop PTS, and severe PTS, is poor.
Elevated D-dimer measured off oral anticoagulants has been found to be predictive of recurrent VTE [6,7], perhaps because D-dimer levels reflect chronic thrombus or ongoing activation of coagulation. The pathophysiology of PTS is thought to relate to a combination of persistent venous obstruction and valvular reflux [8–10]. Whether D-dimer might be a useful marker to predict the development of PTS after DVT is not known. While venous valvular reflux is a common occurrence after DVT, evidence on the relationship between venous valvular reflux and PTS is conflicting [8–13].
In the present study, we sought to determine the association between D-dimer level and the subsequent development of PTS, as well as the association between venous valvular reflux and PTS, in a well-characterized population of patients with objectively diagnosed DVT.
Materials and methods
The Venous Thrombosis Outcomes (VETO) Study was a prospective multicenter cohort study whose primary aims were to determine the frequency and risk factors for the development of PTS in patients with acute symptomatic DVT, the effects of PTS on health-related quality of life and the economic burden of DVT and PTS [14,15]. The present study was a funded substudy of the VETO study.
The methods of the VETO Study are described in detail elsewhere . In brief, the cohort consisted of consecutive patients with acute DVT recruited from emergency departments, outpatient clinics and inpatient wards of eight university-affiliated hospitals in Quebec and Ontario from April 2001 to September 2004. Eligible patients had acute symptomatic DVT of the lower limb that was objectively diagnosed within the previous 7 days. Exclusion criteria included estimated lifespan of < 3 months, inability to read and understand English or French, geographic inaccessibility for follow-up and lack of informed consent. The study was approved by the Ethics committees at all study sites and written informed consent was obtained from the patients prior to their entry into the study. Based on an expected 20% incidence of PTS during follow-up and a projected 25% rate of death or loss to follow-up , a sample size of at least 360 patients was required in order to detect a minimum 2-fold difference in potential risk factors for PTS (such as inherited thrombophilia and proximal location of DVT, expected presence in DVT patients without PTS 20% and 40%, respectively) with 80% power (P = 0.05, two-sided) .
Study visits took place at baseline, 1, 4, 8, 12 and 24 months. At baseline, demographic and clinical data on potential risk factors for PTS were collected. At each follow-up visit, information on the use of anticoagulants and elastic compression stockings was recorded, and a clinical assessment for PTS was performed, as described below. Patients were instructed not to wear compression stockings on the day of their clinic visit; hence assessors were not aware of the use of elastic compression stockings by the patients. All patients who developed symptoms suggestive of VTE during follow-up were evaluated at the study centers to rule in or rule out recurrent VTE using algorithms and diagnostic criteria as previously detailed . All suspected recurrences were adjudicated by the study adjudication committee.
At the 4-month visit, patients had blood drawn for thrombophilia testing and plasma samples were frozen at −80° and reserved for future use . For this substudy, reserved samples were thawed in order to measure D-dimer levels using an enzyme-linked immunosorbent assay (VIDAS D-dimer; BioMérieux, Lyon, France). The coefficient of variation for the VIDAS D-dimer observed in the study laboratory ranged from 2.3% to 6.1% for normal D-dimer levels, and 5.5–8.6% for extreme high D-dimer levels.
At the 12-month visit, patients had a compression ultrasound to detect the absence or presence of reflux at the popliteal vein. Reflux evaluation was standardized in all centers according to a predefined protocol. Absence of venous reflux (i.e. venous competence) was defined by the sudden cessation of the venous Doppler signal after a Valsalva maneuver, or after manual compression of the ipsilateral calf for at least 10 s, followed by sudden release. Venous valvular reflux was considered present when there was retrograde flow through the valve for at least 1 s with either of the above maneuvers . Although not established as a gold standard, this Doppler assessment of venous reflux has been previously shown to have high inter-rater and intra-rater reliability [18,19]. The evaluation of residual vein thrombosis by compression ultrasound was not performed as part of the study protocol, and information on whether this was done as part of local clinical practice was not collected.
Patients were evaluated for PTS using the Villalta scale, which has been shown to be a reliable and valid measure of PTS . At each follow-up visit, patients were asked to report on the five symptoms assessed by the scale, and trained personnel who were kept blinded to the patient’s responses rated each leg on the six signs assessed by the scale. Assessors were unaware of D-dimer or reflux results. For the primary study outcome, a patient was considered to have developed PTS if the Villalta score in the same leg as the index DVT was ≥ 5 on at least two visits starting at the 4-month visit or later, or was ≥ 5 at the final follow-up visit. Severity of PTS was defined as follows: a score of 5–9 represented mild PTS; 10–14, moderate PTS; and ≥ 15 or the presence of a venous ulcer, severe PTS.
The primary aim of the statistical analysis was to assess associations between D-dimer level, valvular reflux and PTS. D-dimer was examined as a continuous variable (units, μg L−1) and as a dichotomous variable (above or below the manufacturer’s cut-off for normal, < 500 μg L−1). Bivariate analyses were performed for the main independent variables (D-dimer, valvular reflux) and dependent variables (PTS, severity of PTS) using chi-square or anova statistics.
Subsequently, logistic regression analysis was performed to determine the associations between D-dimer level, valvular reflux (main independent variables) and the development of PTS, taking into account the following potential confounding variables: warfarin use at time of D-dimer blood draw, age, gender, BMI, history of previous DVT, type of DVT (idiopathic, cancer related or associated with transient risk factor), proximal extent of DVT, use of compression stockings and recurrent VTE during follow-up. Potential confounders were selected a priori then partial and full logistic models were built and compared in terms of crude and adjusted OR for D-dimer and valvular reflux. A P-value of the Wald chi-square of < 0.05 was the threshold used for statistical significance, but P-values < 0.10 were noted.
D-dimer analyses were also performed without and with stratification by warfarin use at the time of D-dimer blood draw, as anticoagulants may affect D-dimer levels.
Finally, bivariate analysis (using t-test and chi-square statistics, and including stratification by anticoagulation at time of D-dimer testing) as well as logistic regression (adjusting for possible confounders) were performed to determine whether D-dimer at 4 months and valvular reflux measured at 12 months were associated. All analyzes were performed using sas version 9.1 (SAS Institute, Cary, NC, USA).
Characteristics of study subjects and frequency of PTS
As previously reported, 1032 patients were screened for participation and of these, 645 had one or more exclusion criteria, leaving 387 participants (37.5%) who were enrolled and followed in this study . During follow-up, recurrent VTE occurred in 31 subjects.
Baseline characteristics of the study cohort are shown in Table 1. Mean duration of warfarin use was 34.1 weeks overall, 26.5 and 38.8 weeks in patients with distal and proximal DVT, respectively, and 39.6, 29.5 and 29.1 weeks in patients with idiopathic, cancer related and transient risk factor DVT, respectively. In all, 52.2% of patients reported using elastic compression stockings during follow-up. Of the 344 who were able to be assessed for PTS (as per our definition that required assessment on at least two follow-up visits after the 4-month visit), PTS developed in 155 (45.1%) patients. Of these, 87 patients (25.3%) had mild PTS, 43 (12.5%) moderate PTS and 25 (7.3%) severe PTS. Of the 68 patients who developed moderate or severe PTS, 38 did so after the 12-month follow-up.
Table 1. Baseline characteristics of study participants*
Entire cohort n = 387
PTS†n = 155
No PTS†n = 189
PTS, post-thrombotic syndrome; BMI, body mass index. *n (%) unless indicated otherwise. †Denominator is the 344 patients who were assessable for PTS, as per our definition requiring at least two consecutive visits . ‡Cancer that was diagnosed within the past 6 months or is undergoing treatment, metastatic or terminal. §Transient risk factors defined as surgery, trauma or immobilization for ≥ 3 days within last 3 months. ¶No cancer or transient risk factors.
BMI (kg m−2): mean (SD)
Chronic obstructive lung disease
Angina or myocardial infarction
Congestive heart failure
Known musculoskeletal condition affecting hip or leg, same side as deep venous thrombosis
Characteristics of index deep venous thrombosis
Proximal deep venous thrombosis
Highest extent, proximal deep venous thrombosis
Superficial femoral vein
Common femoral vein or iliac vein
Distal deep venous thrombosis only
Provoking features of deep venous thrombosis
Transient risk factors§
Concurrent symptomatic pulmonary embolism
Previous deep venous thrombosis
Previous ipsilateral deep venous thrombosis
D-dimer and PTS
Blood samples were available for the 305 patients who consented to have their blood drawn. Of these, 140 (45.9%) developed PTS and 165 (54.1%) did not. Mean (SD) D-dimer level was 712.0 (550.9) μg L−1 in subjects who developed PTS compared with 444.0 (1263.4) μg L−1 in those who did not develop PTS, for a mean difference of 268.0 μg L−1 (95% CI 53.9, 482.1; P = 0.02). Patients with moderate/severe PTS had similar mean D-dimer levels to patients with mild PTS [722.9 (1424.7) and 696.7 (1003.6) μg L−1, respectively; P = 0.90]. At the time of the 4-month blood draw, 213 patients were taking anticoagulants. When analyses were stratified for use of anticoagulants, mean D-dimer levels were significantly higher in subjects who developed PTS vs. those who did not in both strata; however, this difference was greater in patients who were not taking anticoagulants at the time of blood draw compared with those who were on anticoagulants (Fig. 1).
In the study population as a whole, when D-dimer level was dichotomized as positive vs. negative, the OR for developing PTS was 1.13 (95% CI 0.68, 1.88; P = 0.62). However, when the analysis was restricted to subjects who were not on warfarin at the time D-dimer was measured, the OR for the development of PTS if D-dimer was positive was 3.79 (95% CI 1.46, 9.85; P = 0.005).
In logistic regression analyses modeling D-dimer as a predictor of PTS, D-dimer level (continuous variable) was statistically significant in both crude and adjusted models (Table 2). After adjustment for warfarin use at the time of blood draw as well as age (continuous), gender, BMI (continuous), history of previous DVT, type of DVT, extent of DVT and use of compression stockings, D-dimer significantly predicted PTS [per 100 μg L−1 difference in D-dimer, OR 1.05 (95% CI 1.00, 1.09; P = 0.03)]. Adding recurrent VTE during follow-up to the full model did not modify the association between D-dimer and PTS (Table 2).
Table 2. Association between D-dimer and PTS: results of multivariate analysis
Odds ratio for 100 μg L−1 difference in D-dimer (95% CI)
Odds ratio for 1000 μg L−1 difference in D-dimer (95% CI)
*D-dimer used as continuous variable unless otherwise specified. †See Statistical analysis in Methods section. ‡Type of DVT: cancer-related, associated with transient risk factor (vs. idiopathic). §Extent of DVT: popliteal, superficial femoral vein, common femoral vein or iliac vein (vs. calf vein).
1.04 (1.00, 1.07)
1.43 (1.04, 1.96)
D-dimer, warfarin use at time of blood draw
1.05 (1.01, 1.10)
1.71 (1.16, 2.52)
D-dimer, warfarin use at time of blood draw, age, gender, BMI, history of previous DVT, type of DVT‡, extent of DVT§, use of compression stockings, recurrent VTE during follow-up
1.05 (1.01, 1.10)
1.62 (1.05, 2.50)
Odds ratio for unit difference in log D-dimer
1.21 (0.96, 1.53)
Log D-dimer†, warfarin use at time of blood draw, age, gender, BMI, history of previous DVT, type of DVT‡, extent of DVT§, use of compression stockings, recurrent VTE during follow-up
1.33 (0.96, 1.84)
As the distribution of D-dimer is right skewed, we also assessed the association between log transformed D-dimer and PTS. The P-value for log transformed D-dimer in the full model was 0.09, with an OR of 1.33 (95% CI 0.96, 1.84) for a unit difference in log D-dimer. For example, the OR for developing PTS for a D-dimer level of 1000 compared with a D-dimer level of 100, adjusting for other risk factors for PTS, is 1.33.
When we repeated the above logistic regression analyses stratified by warfarin use at the time of blood draw and modeling D-dimer level (continuous variable) as a predictor of PTS, similar findings as above were noted, except that the adjusted odds ratio per unit difference in log D-dimer was 2.33 (95% CI 0.89, 6.10) in those not on warfarin vs. 1.25 (95% CI 0.87, 1.79) in those on warfarin.
Valvular reflux and PTS
Two hundred and thirty-three patients agreed to have a compression ultrasound examination at 12 months for the assessment of reflux at the popliteal vein. Ipsilateral venous valvular reflux was present in 50.9% (59/116) of subjects with PTS compared with 42.7% (50/117) of subjects without PTS (P = 0.21). There was a positive association between presence of reflux and severity of PTS: reflux was detected in 65.3% (32/49) of patients who developed moderate/severe PTS, compared with 40.3% (27/67) of patients who had mild PTS (P = 0.02).
In multivariate modeling adjusted for potential confounding variables, namely, age, gender, BMI, previous ipsilateral DVT, type of DVT (idiopathic, cancer related or associated with a transient risk factor), extent of DVT and use of compression stockings, valvular reflux was independently associated with moderate-to-severe PTS (OR 2.72 for moderate-to-severe PTS compared with no or mild PTS (95% CI 1.25, 5.90; P = 0.01) (Table 3).
Table 3. Association between venous valvular reflux and moderate-to-severe PTS: results of multivariate analysis
OR for presence of valvular reflux (95% CI)
*Type of DVT: cancer-related, associated with transient risk factor (vs. idiopathic). †Extent of DVT: popliteal, superficial femoral vein, common femoral vein or iliac vein (vs. calf vein).
2.61 (1.36, 5.05)
Valvular reflux, age, gender, BMI, previous ipsilateral DVT, type of DVT*, extent of DVT†, use of compression stockings
2.72 (1.25, 5.90)
Association between D-dimer and valvular reflux
In the bivariate analysis, D-dimer at 4 months, whether used as a continuous or dichotomous variable, was not found to be associated with valvular reflux measured at 12 months (t-test and chi-square P-values 0.67 and 0.66, respectively). When stratifying on use of anticoagulation at 4 months, no association was found between D-dimer and valvular reflux. In logistic regression analysis adjusted for warfarin use at time of blood draw, age, gender, BMI, history of previous DVT, type and extent of DVT and use of compression stockings, D-dimer did not predict valvular reflux (P = 0.15).
In this multicenter, prospective cohort study of patients with DVT, we evaluated the association between D-dimer level, venous valvular reflux and PTS. We found that higher D-dimer levels measured 4 months after DVT are predictive of an increased risk of developing PTS, and that venous valvular reflux is associated with moderate-to-severe but not mild PTS.
Measurement of D-dimer is frequently carried out in clinical practice for the diagnosis of acute VTE, and more recently, the use of D-dimer as a predictor of risk of VTE recurrence has been examined [21–23]. However, to our knowledge, the association between D-dimer level and the development of the PTS has been addressed in only one previous report. In a prospective study of 406 patients with DVT, Stain et al.  examined the association between D-dimer (Asserchrom D-dimer ELISA assay; Boehringer Mannheim, Germany), measured 3 weeks after discontinuation of oral anticoagulation, and PTS. The PTS was assessed by the Clinical-Etiologic-Anatomic-Pathophysiologic (CEAP) classification which considers signs of PTS only for scoring as opposed to the Villalta scale which takes into account both symptoms and signs. After adjusting for age, gender, proximal DVT, BMI and factor V Leiden, the OR for PTS associated with a D-dimer > 500 ng mL−1 was 1.9 (95% CI 1.0, 3.9).
Although our study population differed from the population in the study by Stain et al. (e.g. our study included cancer patients and those with recurrent VTE), the risk of developing PTS was similar in both studies (45.1% vs. 43.3%, respectively). Contrary to Stain’s study, we assessed the association between D-dimer and PTS both on and off oral anticoagulation, and examined D-dimer both as a continuous and dichotomous variable. When we dichotomized D-dimer, our finding of an OR of 3.8 for PTS among patients not taking anticoagulants was higher than the OR of 1.9 reported by Stain. This difference in ORs may have been because of the differences in timing of D-dimer assessment in the two studies, as well as the different D-dimer assays and definitions of PTS used.
In our study, about two-thirds of patients were receiving anticoagulation at the time of D-dimer measurement. While D-dimer levels were lower among patients on warfarin than those off warfarin, the association with PTS was similar in both patient groups and results of the logistic regression analysis showed that the strength of the association between D-dimer level and PTS did not change appreciably when adjusting for anticoagulation, although analyses stratified by use of warfarin suggest that D-dimer levels off warfarin may be a more useful predictor of PTS than levels measured while on warfarin. An elevated D-dimer level may be an intrinsic marker of residual thrombus and/or persistent activation of clotting or inflammatory pathways that could increase the propensity to develop PTS. When D-dimer was dichotomized (cut-off 500 μg L−1), an association between D-dimer and PTS among patients on warfarin was not found. Patients taking anticoagulants have lower D-dimer levels, thus limiting the usefulness of a qualitative D-dimer value, as is frequently used in practice, as a predictor of PTS in this population. The actual value of D-dimer in these patients may be clinically more useful.
We measured D-dimer levels 4 months after DVT diagnosis because that was the time point of the blood draw for thrombophilia testing in the VETO Study. However, this time point may be sensible clinically, as measurements made earlier could have reflected resolving thrombus from the initial DVT. For example, Kuruvilla and colleagues  studied D-dimer levels in patients with VTE who were receiving anticoagulation and found that more than 85% of patients had a negative D-dimer (cut-off 200 μg L−1, tested using the Instrumentation Laboratory method on an Amelung Coagulometer; Sigma Diagnostics, St Louis, MO, USA) by 3 months after the VTE event. On the other hand, at 1 month post-VTE, 39% still had a positive D-dimer.
We found that venous reflux measured at the popliteal vein at 12 months after DVT was more prevalent in patients with moderate-to-severe PTS than in those with mild PTS, and reflux was significantly associated with moderate-to-severe PTS in adjusted analyses. Results of previous studies of the association between venous valvular reflux and PTS have been conflicting. In one study , venous reflux was assessed in 82 patients 7–13 years after a DVT; 64% of patients with severe PTS had both deep and superficial vein reflux. Conversely, the same authors later conducted a 2-year follow-up study on a different study sample  and found that superficial vein reflux was only weakly associated with PTS. Two other studies [8,10] showed that a combination of residual vein thrombosis and venous reflux increased the risk of developing PTS. However, in a study by Roumen-Klappe et al.  of 93 patients examined up to 6 years after DVT, venous reflux measured in the popliteal vein was not a predictor of PTS. In view of the disparate evidence, it remains unclear whether the development of venous reflux after DVT is a risk factor for the development of PTS or of severe PTS, is a marker of PTS itself, or occurs as a consequence of severe PTS. Further mechanistic studies are needed to determine the temporal associations between reflux, venous obstruction and the development of PTS.
Strengths of the present study include that our cohort was comprised of consecutive patients with DVT whose baseline characteristics were carefully documented, assessment for PTS was performed by evaluators who were unaware of D-dimer or reflux results and reflux was assessed by evaluators who were unaware of subjects’ PTS status. Also, we examined the effect of D-dimer levels both on and off anticoagulation, thus taking into consideration the known effect of anticoagulation on D-dimer value [25,26], and we examined D-dimer as both a continuous and dichotomous variable, as the actual quantitative value of D-dimer may be more informative regarding ongoing activation of coagulation and risk of PTS. Finally, we took into account in our analyses known risk factors for PTS.
The present study has several potential limitations. Not all patients who participated in the VETO cohort agreed to provide blood samples or have the 12-month ultrasound for reflux. However, the proportion of subjects who developed PTS was similar in those who did (45.9%) and did not (38.5%) provide blood samples. As D-dimer and venous reflux measurements were carried out only once during the study, we are unable to define a sequence for events or establish a cause and effect relationship between D-dimer and PTS or reflux and PTS. We cannot comment on whether the same results would have been obtained with other D-dimer assays. In addition, as we followed patients for 2 years after DVT, patients who developed PTS beyond that time period may have been missed. Finally, we did not have data available on quality of anticoagulation (whether INR was therapeutic) at the time D-dimer was measured.
In conclusion, measurement of D-dimer 4 months after DVT, particularly in patients not taking anticoagulants, may help predict which patients are at higher risk of developing PTS and who thus may be more likely to benefit from preventive strategies such as compression stockings for PTS. Further prospective studies of the potential association between D-dimer and PTS are warranted. If confirmed, clinical risk prediction algorithms that include D-dimer to predict the risk of PTS should be developed and validated. Finally, venous valvular reflux was found to be strongly associated with moderate-to-severe PTS. Further studies are needed to evaluate whether detection of venous reflux after DVT has a role in determining which patients are at subsequent risk of more severe forms of PTS.
Study conception and design: S. Desmarais, S. R. Kahn, J. Latella. Analysis and interpretation of the data: J. Latella, S. R. Kahn. Drafting of the article: J. Latella, S. R. Kahn. Critical revision of the article for important intellectual content: All authors. Final approval of the article: All authors. Provision of patients, collection and assembly of data: S. R. Kahn, M-J. Miron, A. Roussin, S. Desmarais, F. Joyal, J. Kassis, S. Solymoss, L. Desjardins, J. S. Ginsberg. Obtaining of funding: S. R. Kahn.
The authors would like to thank the VETO study personnel as well as the patients who participated in the VETO study. This substudy to the VEnous Thrombosis Outcomes (VETO) Study was funded by the Cardiovascular Research Network of the Fonds de la Recherche en Santé du Québec (FRSQ). The VETO Study was funded by the FRSQ and by an unrestricted grant-in-aid from GlaxoSmithKline. J.S. Ginsberg is a recipient of a Career Award from the Heart and Stroke Foundation of Ontario and holds the David Braley and Nancy Gordon Chair in Thromboembolic Disease at McMaster University. S.R. Kahn is a recipient of a National Investigator Award from the FRSQ.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.